The listing's verdict capsule now carries white text on muted semantic
fills instead of the neutral secondary tint: Saves #45825E, Costs
#A9635B, Break even #707070, Closest #58789D.
Fixed tones, not translucent tints, because white text on a 15% tint
over a white row measures ~1.8:1. Each tone sits at a relative
luminance of ~0.18 — the band where white clears 4.5:1 AND the pill
still separates ~3:1 from both a white row and the dark card fill — so
one value each works in light and dark without a dynamic colour. The
saturated palette colours cannot carry white text at all: systemGreen
measures 2.2:1, systemRed 3.6:1.
"Save" becomes "Saves", matching the detail sheet and reading as a
statement of the outcome; amounts, symbols, padding, radius and the
shape-follows-line-count rule are unchanged. Saves and Costs share an
84pt minimum width so their trailing edges line up down the list.
The price and its capsule are now one vertically grouped element (5pt),
with the badge keeping the column's full 12pt above, so the row no
longer reads as three equal lines.
Station names cap at three lines with a tail ellipsis: unbounded wrap
let one very long forecourt name set the height of the whole row. 59 of
the 8,073 names in the dump exceed three lines and now ellipsise; the
full name is always in Station Details. Addresses still wrap freely.
The baseline pill moves onto ListingFuelSavingsCapsule so the listing
shows the same soft blue and white text as its siblings, leaving
FuelSavingsCapsule to the detail sheet, which it now alone uses.
Badges keep their tinted styling: they annotate a price rather than
report an outcome.
Verified: 261/261 swift test; Debug simulator and Release device
builds; all four fills and their white text measured from simulator
pixels in both appearances, and the detail sheet's .blue absent from
the listing entirely.
Station and Favourite rows share one capsule, so both now read
"Save / £3.10" and "Costs / £1.20 more" instead of the amount-free
"Saves" / "Costs more". "(approx.)" is appended on its own line when the
journey behind the figure was estimated. Break even carries no figure —
±£0.05 is noise — so it stays a single line unless the qualifier adds one.
- Split `listingCapsuleHeadline()` into a structured `ListingVerdict`
(headline / amount / qualifier) drawn as one Text per line, so where the
capsule breaks is decided by the data rather than by the width the price
column happened to hand it: the figure can never straddle two lines.
- The capsule's shape follows its line count: one line stays a Capsule, two
or more switch to a rounded rect at 8pt, whose corners do not crowd a
taller block the way a pill's semicircular ends do.
- The Key sheet's legend renders the same component with `showsAmount:
false`, so it demonstrates the wording without displaying the placeholder
£1.00 figure it was built from.
- The new words ("Save", "Costs", "(approx.)", "%@ more") go into
Localizable.strings so the copy editor owns them; amounts run through
FuelSavingsFormat.money and the amount line is monospaced.
- VoiceOver keeps the single-line form, since a drawn break is a pause
mid-phrase when spoken.
Tests cover the line structure for all four statuses, the estimated
qualifier and the amount-free legend. Pixel measurement on the simulator
confirms the shape rule: one line measures a radius of h/2 (a pill), two
lines measure 8pt, in both light and dark.
- Alerts: drop the alert-type picker and the Best Value alert path, so
proximity alerts always follow the cheapest station. A stored `bestValue`
selection now falls back to cheapest when loaded.
- Key sheet: the price-rating, savings, station-label and favourite wording
comes from `KeySheetContent` in Shared/FuelSavings.swift — a Foundation-only
model with tests that pin the exact meaning of each indicator — and the
sheet is built from that model rather than inline strings.
- BACKLOG: record the plan to restore Best Value proximity alerts behind one
shared evaluator (user-location baseline, a single disclosed qualification
rule, and real/Test alert parity).
- Bump CURRENT_PROJECT_VERSION to 29.
The Settings sliders called persistFuelSavingsSettings() on every value change,
which rewrote all four keys — four UserDefaults writes and four keychain round
trips per step, on the main thread, up to 90 times per drag. That is what made
them stutter.
saveFuelSavingsField(_:from:) writes just the field that moved, so a slider
gesture touches one key.
writeKeychain now uses SecItemUpdate and only falls back to SecItemAdd on
errSecItemNotFound. Delete-then-add cost two round trips to securityd and
briefly left the key holding no value at all — a window where the keychain, the
authoritative mirror on read, could disagree with the app-group defaults.
Tests pin the store half: a single-field save leaves the other three keys alone,
and the whole-settings save still writes all four.
The sheet leads with the station name and address inside the card, above the map, then a two-column facts row (fuel price on the left, distance and return on the right, split by a hairline) and the two station actions divided the same way. The Best Value block becomes a tinted card: verdict badge, the two figures the estimate is made of, the working and the total. "How Best Value works" moves to its own row below it, using the shared HowBestValueRow the Key sheet already shows.
Presentation only: every figure still comes from the shared sources (FuelStore.priceText, stationDistanceDisplay, FuelSavingsResult), so the sheet cannot disagree with the listing row, the widget or the Live Activity, and no calculation, ranking or freshness rule changed.
StationDistanceDisplay.returnLabel(unit:) and FuelSavingsFormat.extraTravelDistanceLabel(...) are split out of their caption forms so a surface printing the return or extra-travel figure on its own line takes it from the same place, confidence marker included; their existing output is unchanged, pinned by tests. FuelSavingsSummaryView is deleted - the sheet's card supersedes it and nothing referenced it.
DEBUG-only QA hooks: -qaConfigureSavings configures a Fuel Savings plan (the Best Value card and the Worth the Drive sort are gated on it), -qaDetailBottom scrolls to that card (defaultScrollAnchor is a no-op on a List) and -qaCopyPostcode holds the copied confirmation long enough to screenshot.
Two columns split by a 1pt hairline: identity on the left at a fixed 52%
share so rows line up when scrolling, price and savings on the right,
leading-aligned on one axis with the price dominant. The favourites star
sits under the station logo, freeing the width the pounds-and-pence price
needs to render in full rather than scaled or truncated.
Badges stay tinted rather than filled — BEST VALUE green, CHEAPEST blue,
none for nearest — and the verdict capsule is one shared view across the
listing, the detail sheet and the compact summary, so wording and colour
cannot drift between them.
Best Value's minimum saving moves from Alerts to Settings, Fuel Savings,
beside the fill size and MPG that produce the saving; the Alerts tab reads
it back instead of offering it twice. Its options are now 50p steps to £5
(was 50p/£1/£2/£5), so £2.50 and £3 are reachable, with a test pinning the
step, the ceiling and the store's string round-trip.
The verdict capsule scales to 0.7 before truncating. Measured on a 393pt
screen its copy is handed ~69pt while "Closest station" needs ~90pt, so the
old 0.8 floor could not reach it and rows showed "Closest stati..." instead.
Verified: Debug build clean, swift test 249/249.
The relay's price_updated is the maximum price_last_updated across the grades it maps, so it cannot vouch for the selected grade: stop using it to score, gate or warn about Best Value.
- FuelPriceFreshness is now station-level and display-only: one age per station, fresh/outdated/unknown/invalid, "May be outdated" past 30 days, with future-tolerance and non-finite guards.
- Best Value restores the closest-station-selling-the-selected-fuel baseline and its savings ranking; eligibility gating, eligible-first sorting and the unavailable banner are gone.
- Price age appears in the More info sheet only - no listing-row warnings, no extra banners.
- Copy reverted from "closest eligible station" to "closest station"; stale price-confidence strings removed, "May be outdated" added.
- Dormant per-grade timestamp plumbing removed, and regression tests added for every fuel type, missing/invalid/future stamps and the bundled offline dump.
Dated pricing had crowded More info and the station rows, and most of the
extra copy described a distinction the data cannot make: every station
carries a single station-wide timestamp, so "grade-specific vs
station-wide" never varied in practice.
- More info shows one row, "Price updated 28h ago" (or "Not provided"),
replacing the separate station-report and fuel-update-time rows and the
blanket "price freshness is not confirmed" label.
- Rows and the detail sheet stay quiet when a saving is shown; only an
ineligible comparison explains itself. Provisional savings already read
as estimates from the capsule, so the fuel-named warning lines are gone.
- Freshness/eligibility logic is unchanged: it still gates the baseline and
still supports grade-specific timestamps for when the relay emits them.
- Tests updated; localisation keys for the removed copy dropped.
- Add FuelPriceFreshness: selected-fuel-aware eligibility plus a shared
comparison confidence, so a saving is presented as verified only when
both the candidate and the baseline pass the gate for the fuel on screen.
- Rank against the closest station with an eligible selected-fuel price
instead of the physically closest station, so a single stale neighbour
can no longer invalidate every comparison in the list. Ordinary Closest
sorting still includes every priced station.
- Keep ineligible prices visible, fall back to pump-price order, and label
Best Value unavailable when no comparison qualifies.
- Widget: same baseline and confidence path as the app.
- Tests: stale-closest baseline, alert baseline, eligibility policy,
boundary and missing-timestamp regressions.
Distances (the "Approx. 1.5 mi" report):
- Refuse junk CoreLocation fixes (invalid accuracy, or a coarse >300 m fix
within 3 km of the position already held; a coarse fix in another city is
real travel and is still accepted).
- Resolve ONE display basis per render through FuelStore.distanceDisplayOrigin,
used by rows, radius filter, sorting, savings figures and the Live Activity,
so a stray fix can no longer re-base the list outside the route cache's 600 m
ring while the routed values sit unused in the cache.
- Treat an incomplete route set as a state: retry it after 30 s rather than the
full 10-minute throttle, and repair a stale incomplete flag once every
in-radius station has a route (nothing left to route, flag stuck down).
- Widget extension resolves the same origin for fetch, radius, ordering and
labels, and keeps its own fix in fuelboard.widgetLocation instead of
overwriting the app's GPS fix.
Alerts:
- Best Value as a distinct alert mode on the existing geofence/notification
pipeline, with a persisted threshold; reverts to cheapest when the feature is
unconfigured and is not selectable while disabled.
Diagnostics (DEBUG, gated): FB_DIST_FOCUS now prints FB-DIST-ORIGIN (candidate,
accuracy, anchor, offset, resolved branch, served origin), FB_FAKE_FIX feeds a
synthetic fix through the real guard, FB_STALE_INCOMPLETE forces the stale-flag
state, and the route-fill status is surfaced in Settings behind debug mode.
Distance confidence
- Every distance surface takes the figure AND its "Approx." marker from one
FuelStore.stationDistanceDisplay call (station rows, detail sheet, widget,
Live Activity, watch, Siri), so a surface can never mark one figure approximate
while showing a routed one. Radius filters and Closest ordering test that same
displayed distance, so no row can read a distance outside the radius that
filtered it (the reported "14.6 mi inside a 10-mile list").
Accurate distances (Settings -> Distances, default off)
- Off: the browse layer serves the straight line, marked Approx., for every row,
so two installs agree to the metre and the app makes no route requests at all.
- On: routed distances for the whole search area, unmarked, served only from a
COMPLETE set anchored within 600 m — a partial fill reverts every row at once
rather than mixing road and straight-line bases.
- Best Value maths sits deliberately outside the toggle: it keeps using measured
road routes and its own confidence marking, so money figures never move with
the switch.
- Fills run only while the toggle is on and the app is frontmost; capacity raised
to 600 stations per pass; completeness tolerates pins Apple Maps has no route
for, which otherwise left an area permanently incomplete and the whole list
stuck on estimates.
Performance (the reported slowdown switching to Closest / launching with it)
- FuelStore.accurateDistancesEnabled was a UserDefaults(suiteName:) read per
call: a synchronous cfprefsd round-trip measured at ~1.9 us, ~200k calls per
Closest sort. Memoised, with the setter keeping it in step.
- Both keyed sorts decorate their keys — one distance per station, carried as an
index instead of copying whole stations into the decorated array — and the
comparison baseline plus per-station journeys now come from ONE pass instead of
two full scans per tab. The Stations tab also reuses its single sorted list for
the count and the TOP badge instead of sorting the country three times.
- Measured on iPhone SE (7,914-station Closest pool): sort 36.6 ms -> 17.1 ms per
list, Cheapest 6 ms, Best Value 11 ms.
- DEBUG probes: FB_PERF list timing, FB_PERF_BENCH toggle-read benchmark, and
FB_DIST_FOCUS now reports the served value, marker and mode (and re-prints when
a fill lands). FB_ACCURATE_DISTANCES / FB_SORT_MODE env forms for devices, where
devicectl eats a leading-dash argument.
Tests: 198, 0 failures. Device and simulator Debug builds green.
Users who finished onboarding before Best Value existed are "completed, not
fresh" forever, so the first-run gate never fires again for them and the feature
stayed invisible unless they went looking in Settings. They now get the Best
Value page on its own, once per announcement version — deliberately NOT a replay
of the onboarding flow, which would re-run permission pages they answered long
ago and has no skip.
- FuelStore: "fuelboard.bestValueIntroVersion", stored through the same
saveString/loadString path as the onboarding flag and compared against a
CODE-SIDE "currentBestValueIntroVersion". Deliberately not
CFBundleShortVersionString: keying on the bundle version would re-announce on
every bugfix release. Bump the constant to re-announce on purpose.
- FuelStore.shouldAnnounceBestValue(settings:hasSeen:onboardingDue:isSuppressed:)
holds the rule, with every fact injectable so it is unit-tested instead of
being driven through a simulator. It declines when onboarding is due or on
screen, when this announcement version has been seen, and when the user has
already configured Best Value themselves.
- BestValueAnnouncementHost wraps the app's content and owns the presentation
state, so ContentView's change is a single wrapper and the whole feature lives
in the new file. It reloads the stored settings on dismiss, because the page
writes straight to the store rather than through ContentView's binding.
- The page is now ONE view, BestValueIntroContent, used by first-run onboarding
(page 5 of 6) and by the announcement, so the copy, the sliders and the
"declining is a decision" rule cannot drift. OnboardingView drops its private
duplicate of the page, its bindings and its read-outs, and finish() marks the
announcement seen so a new user never sees the page twice.
QA hooks (DEBUG): -showBestValueIntro forces the announcement, and -skipOnboarding
suppresses it. Worth knowing now that a returning install announces on a PLAIN
launch, so tab screenshots need -skipOnboarding.
Verified on the simulator (OCR of the rendered screens): a plain launch of an
install that has completed onboarding shows the single announcement page — Best
Value title, explanation, "Turn on Best Value", the "Settings, under Fuel
Savings" note and a "Not now" button, with no page dots; -skipOnboarding shows
no cover at all; -showBestValueIntro forces it. 180 tests, 0 failures (8 new).
Following the sort-segment rename, the rest of the user-facing "Worth the Drive"
copy moves to "Best Value".
- Settings: the toggle is now just "Best Value" (was "Enable Worth the Drive"),
and the narrative footer opens "Best Value estimates fuel cost only, …".
- More info sheet: the section header is "Best Value" (was "Worth the drive?"),
and the section itself is now conditional. `fuelSavings` is nil whenever the
feature is off — `savingsLookup` returns nil when it is not configured — so
the heading is never left dangling over an empty section while the feature is
disabled.
- The unavailable text reads "Best Value unavailable: …".
- Comments and doc comments that named the sort were updated so they match the
UI instead of contradicting it.
Deliberately unchanged: the enum's raw value stays `worthTheDrive`, so stored
sort preferences and widget configs keep decoding, and FuelSavingsStatus keeps
its verdict wording ("Worth the drive" / "Not worth the drive") — that describes
a station's verdict rather than the feature, and is not shown in the UI.
Verified on the simulator: with the feature OFF the more info sheet shows
Address / Distance / Directions and no heading at all; with it ON the sheet
shows the "Best Value" section; Settings reads "Fuel Savings" / "Best Value"
with no "Enable" anywhere. 172 tests, 0 failures.
The third sort segment, the caption line it drives on the Stations tab, and the
Settings sentence about sorting now read "Best Value" — that is
SortMode.worthTheDrive.displayName plus the two copy sites that name the sort.
The feature that backs it keeps the name "Worth the Drive": the Settings toggle
("Enable Worth the Drive"), the Fuel Savings section header and footer, the sheet
header "Worth the drive?", the "Worth the Drive unavailable: …" text and
FuelSavingsStatus.worthTheDrive's own wording are all unchanged.
The enum's raw value stays `worthTheDrive`, so stored sort preferences and widget
configs keep decoding — only the label moved.
3 new tests pin the three sort names, the stability of the raw values, and the
fact that the feature's own wording did not change (172 total, 0 failures).
Verified on the simulator: the Stations segmented control reads
Closest | Cheapest | Best Value.
Two things the listing row was saying twice, or too loosely.
- The verdict capsule used a Label, and a Label lays its icon and title out
across the width it is given: in a listing row that left 22pt of empty space
between the symbol and "Save £1.07 after travel". It is now an HStack with
an explicit 3pt spacing (5pt in the detail sheet), so the text sits beside its
icon and the capsule background hugs the pair. Measured on the same row: the
capsule text moved x 412 -> 341 and the pill 270-831 -> 270-691.
- The address line printed the station name a second time. The official Fuel
Finder address field usually opens with the station's own name ("CHIDSWELL
SERVICE STATION, LEEDS ROAD, DEWSBURY, WF12 7QW" for "Jet Chidswell"): 14% of
the dump's addresses start with the exact name, and 13% also end with the
postcode the row appended again. StationAddressLine (new, in Shared/, so the
row and the detail sheet share one rule) drops a leading repeat of the name
and writes the postcode exactly once.
The rules are deliberately conservative: a leading segment carrying a house
number is a real street address and is left alone ("41 ANNACLOY RD ANNACLOY"),
and a partial-word match ("Jet" against "JETSON FILLING STATION") never cuts
into a word. 14 new tests (169 total, 0 failures) pin the shapes found in the
real dump as well as the examples above.
Verified on the simulator: the first listing row's second line went from
"CHIDSWELL SERVICE STA..." to "LEEDS ROAD, DEWSBURY...", and the detail sheet
reads "Sainsbury's Halifax / WADE STREET, HX1 1SB".
The two Fuel Savings figures and the in-list fill-up control were text
fields. A typed figure could sit outside the slider's own bounds, so the
control and the number it claimed to show could disagree, and the decimal
pad needed a Done button to escape.
- Settings: "Average fuel economy" and "Typical fill-up" are now sliders
with a read-out ("45 MPG", "£40" / "40 L") in place of the field. The
warning copy reads "Set a realistic..." rather than "Enter...".
- FuelSavingsAdjustmentView (shared by the Stations Worth-the-Drive header
and the Favourites tab) drops its text field and gains the same read-out;
the typed-value machinery and the -focusFillAmount QA hook go with it.
- Slider bounds live in FuelSavingsCalculator (fillAmountSliderRange/Step,
mpgSliderRange/Step) so the Settings control and the in-list control can
never offer different ranges, and the fill-up slider range stays inside
the calculator's validity ceiling (asserted by test).
- Two tests pin those invariants: every slider position is a valid amount,
and both MPG bounds are valid MPG values. 155 tests, 0 failures.
- Comments that described the removed fields were updated; the keyboard
Done-bar plumbing is kept (no numeric field remains in the app, so it can
no longer fire) rather than deleted, so any field added later still has an
escape route.
Verified on the simulator: Settings shows "Average fuel economy / 45 MPG"
with no field, and the Worth-the-Drive fill-up control shows the unit picker
with a "£40" read-out and a slider. The Settings "Typical fill-up" row sits
below the fold on every simulator available here (and synthetic input cannot
scroll it), so those pixels are unverified.
Calculation audit (brief config: 30 MPG imperial, £60 fill-up, baseline
Rontec Shelf 171.9p at 0.4 mi):
- One mile<->kilometre constant pair (kilometresPerMile 1.609344,
metresPerMile 1609.344) replaces the mixed 1609.34 / 0.621371 / 1.60934
used in the savings path. Effect ~2 ppm (invisible), but the conversions
are now auditable in one place.
- New FuelSavingsCalculationAuditTests (20 deterministic tests, no live
MKDirections) pins the brief's three stations - Tesco 170.9p/2.0 mi,
MFG Hare & Hounds 172.9p/2.8 mi, PGG Napoleon 171.9p/2.8 mi - plus
imperial-vs-US gallon, MPG and budget scaling, the break-even tolerance
boundary, the nearer-candidate clamp, route-unavailable, mixed-basis
estimates and display-rounding boundaries.
- Formula unchanged, and verified correct: fixed quantity (budget / baseline
price), pump saving, round-trip extra distance clamped at zero, travel fuel
priced at the candidate station, +/-5p break even, baseline labelled
"Closest station", sorting on the unrounded net saving. The brief's "£0.18"
for Tesco is the rounded 1 dp extra-distance caption (2.02-2.05 mi reads as
"2.0 mi"), not a rounding bug in the money.
UI (uncommitted since d7a0ecb):
- Done bar drawn in the bottom safe area from keyboard notifications:
.toolbar(placement: .keyboard) is device-dependent (present on the 15 Pro,
absent on the SE on the same iOS build).
- Favourite star in the More-info heading, wired to the same fuel-scoped
toggle as the listing rows at all three call sites.
- CHEAPEST/TOP badge moved above the price so it reads as part of it.
- Listing rows carry the real session fill-up amount instead of a hard zero
(the compact banner does not print it today; a future row layout would).
153 tests, 0 failures. Simulator and device builds succeed.
Optional Fuel Savings feature (off by default): vehicle MPG in UK imperial gallons, fill amount in pounds or litres, persisted through the existing keychain/app-group store. Ranks the WHOLE station pool by estimated net saving (fuel-price saving at each pump minus the extra return-trip driving cost) using MapKit road distances with a straight-line fallback marked (est.).
The closest station with a valid price for the selected fuel is the comparison baseline and is labelled "Closest station"; other rows read "Save £X.XX after travel" / "£X.XX worse off after travel" / "Break even". Labels live in the shared model so they are unit-tested. The TOP badge still marks the cheapest pump price and reads "CHEAPEST" when the list is ranked by net saving.
Road routing now prioritises the Worth-the-Drive candidates plus the baseline and merges cache entries instead of replacing them, so far-out stations get real routes rather than permanent straight-line estimates. Numeric fields gain a Done button above the keypad (decimalPad has no Return key); the fill-up control is hidden unless the feature is configured; Favourites shares one session fill amount with the station list. 131 tests.
The road-distance cache was keyed purely by station ID with no coordinate, so
a value routed to one pin could be served for a same-ID station whose stored
coordinate came from a different source (live fetch vs bundled offline dump
vs a corrected pin), reproducing Maps mismatches.
- RoadDistanceCache now stores each routed pin (CachedRoadDistance{meters,
lat, lng}) and roadDistanceMeters() only serves a value when the displayed
station's coordinate matches the pinned one (within ~11 m).
- Kept the existing origin-distance staleness guard.
- Verified embedded sample data (ids y1/se1/...) never collides with real
relay IDs, and the widget STRICT radius filter drops far-offline samples, so
no actual leak existed in practice — this closes the theoretical stale-pin
channel and future-proofs against coordinate fixes.
The keychain-first completed flag persists across reinstalls (same bundle id),
so a new install was treated as already-completed and onboarding silently
skipped while permissions still prompted. Add install identity:
- Store an install id in BOTH the app-own container defaults (wiped on
reinstall, persists across launches) and keychain (survives reinstall).
- isFreshInstall() is true on first-ever launch or when the local id is
missing/different from keychain (i.e. reinstalled); it seeds a fresh common
id so the next launch of the same install is not 'fresh' again.
- shouldShowOnboarding() = not completed || fresh install -> onboarding now
replays once per fresh/reinstall, never on ordinary launches.
Works on free SideStore (no app-group), using the app's own defaults domain
as the same-install signal.
Stations past the nearest-12 cutoff silently fell back to straight-line,
which explained the remaining app-vs-Apple-Maps gaps (e.g. 3.5 straight-line
vs 5.9 road). Raise the per-pass cap to 40 and route every station within a
25 km straight-line radius (covers the 15 mi max search window) so any station
the widget/Live Activity/list can show gets a real road distance.
Also harden the new cache tests against persistent keychain state across
invocations (reset in setUp) so they pass deterministically.
The widget/Live Activity showed straight-line haversine distance (0.8 mi)
while Apple Maps routes 1.8 mi. Road routing is too heavy for the widget
execution + ~40-70/day refresh budget, so the APP now computes it:
- New RoadDistanceService (app target): for the nearest 12 stations each
pass, calls MapKit MKDirections (free, no API key, matches Apple Maps)
and caches metres keyed by station ID.
- Cache stored in KEYCHAIN (fuelboard.roadDistances) so the widget reads it
even on free SideStore accounts with no app-group container; only valid
within 600 m of the location it was built from.
- Throttled: recompute max every 10 min, or when the user moves > 400 m;
wired into the location-update hook + location onChange.
- Widget face, app station rows + Live Activity show the cached road
distance, falling back to straight-line when absent.
- 4 new cache tests (105 total).
The no-network last resort (bundled real dump) now shows a tappable banner
above the tabs — 'Offline data from 15 Aug / Pull to refresh to update
prices' — so stale snapshot data is never presented as live. Banner date
comes from the dump's own data_updated stamp (FuelStore.offlineDataLabel,
ISO 8601 with/without fractional seconds). Cleared on any successful fetch.
- BundledDumpProvider.dataUpdatedStamp (JSONSerialization, app target)
- ContentView: offlineDataDate state, offlineBanner view (tip-card styling,
orange wifi.slash + arrow.clockwise, tap = force refresh), refresh()
clears/sets it, -forceOfflineDump QA launch hook (serves the dump and
skips auto-refresh so the banner stays up for captures)
- Localizable.strings: 3 keys (banner title/subtitle/a11y)
- +3 tests: OfflineDataLabelTests (101 total)
Legend rows now show each station's average over the days shown
in brackets (pence; signed +X.Xp in vs-cheapest mode, matching the
list-row deltas). Footer under the key explains the figure per mode.
averagePence helper (nil for empty series) + 3 tests; -showTrends
becomes a permanent QA hook like -showKeySheet.
The LAN relay is unreachable from any phone outside the developer's LAN, so
for App Store users a relay fallback was dead weight AND harmful: attempting
the private IP fires the iOS Local Network prompt on a stranger's phone and
adds a timeout before the bundled dump. GitHub raw is reachable anywhere the
relay would be, and more reliably.
- LiveChainProvider: GitHub mirror first EVERYWHERE (focused alert path no
longer relay-first); LAN relay only joins behind the hidden dev flag
fuelboard.relayFallback (Settings → Debug, keychain-first, off by default)
- MirrorFuelProvider: focused() projection replaces the relay's server-side
radius endpoint (fuel filter + within-radius + nearest-first + limit 500)
— the alert path re-checks radius/fuel itself, so behaviour is identical
- Widget: fetchFocused/fetchFuelOnly now GitHub-first via the chain (day-cache
shared through the app group); relay + widget-diag GET gated on the same
dev flag — off-LAN widgets no longer attempt local network at all
- FuelBeacon: fires only when the dev flag is on — consumers make zero
local-network attempts; /stats app-hit spike = GitHub down + dev flag on
- Onboarding: Local Network machinery removed entirely (no probe, no prompt,
no 12 s stall) — data page is now informational: prices download from the
internet, no permissions needed. Replay on cellular can no longer hang.
- Info.plist: NSLocalNetworkUsageDescription copy → optional dev-only wording
- 95 tests (6 new: relay-skip-when-disabled x2, focused prefer-mirror, mirror
focused projection x3, beacon dev-flag gate)
- MirrorFuelProvider: fetches CURRENT prices from raw.githubusercontent
(latest.json pointer → history/<day>.json full dump), app-group day-cache
so the ~2.8 MB dump is re-downloaded only when the mirror pushes a new day
- LiveChainProvider: full path (app refresh) GitHub → relay; focused path
(background alert checks) relay-first so alerts never pull the full dump
over mobile data; records which leg served for About + telemetry
- FuelBeacon: fire-and-forget X-Client app ping to the relay's existing
widget-diag route (zero relay changes) — attribution + cadence on-LAN,
silent skip off-LAN is the reachability datum; /stats app-hit spike =
GitHub path failing
- Bundled dump: FuelBoardDump dataset (real 2026-08-15 snapshot, 8,022
stations) + BundledDumpProvider + scripts/refresh_bundled_dump.sh —
no-network last resort replaces the demo sample set
- ContentView: About meta from the chain; catch falls back cached → bundled
→ sample
- 89 tests (6 new: chain order x4, cache decision, beacon URL); Release
build green; beacon route verified against the live relay (204 + logged)
- Shared/FuelHistory.swift: GitHub mirror read path (latest.json pointer +
history/YYYY-MM-DD.json day files); day math (UTC noon keys), slim day decode
with the shared price band, series + deltaSeries (vs-cheapest rebase),
favourites-only app-group cache (90-day prune), parallel per-day fetches,
404/network = graph gap never error.
- FuelPriceProvider: shared priceBand + mapGrades (single source for live +
history decoding).
- FuelBoard/TrendsView.swift: line chart in the Favourites tab via toolbar
button + sheet; fuel capsule (fuels with favourites only), 7/30/90 range,
Price/vs-cheapest toggle, per-station legend, price-display style on y-axis,
empty/loading/retry states with honest copy; no widget in v1.
- Localizable.strings: Trends keys.
- 83 tests (13 history + URL regression): day math, band guard, series gaps,
delta rebasing, prune, and the appendingPathComponent regression (relative
URL resolution dropped /main — all fetches 404'd).
- Sim-verified: favourites rows + TOP/deltas; sheet controls + building-up
state with live first-snapshot date (archive has 1 day; lines render once
>=2 snapshots).
- Backlog: P1 Trends DONE (unmerged); P0 REMAINING = live provider chain +
telemetry beacon.
- Relay envelope now carries source (api|csv), stations_count, and
data_updated (freshest price_last_updated the GOV.UK server reports,
independent of relay sync time).
- App decodes the new envelope metadata (RelayMeta), persists it after
each full fetch, and shows it in Settings → About → Data source:
Connection (API/CSV), Stations count, Data updated (local-formatted).
- Older relays without the fields decode cleanly (nil meta → em-dash).
- 2 new tests for meta decode + absent-meta tolerance; 37/37 pass.
When several stations share the cheapest price within the radius (within
0.01p), the alert registers a dynamic UNNotificationCategory with up to
4 action buttons (brand + distance, nearest first) and taps open
directions to the chosen station. Shared FuelStore.tiedStations helper
used by both the live geofence path and the Settings real-data test;
35 tests pass.
The proximity monitor was tied to the Stations-tab fuel selection, so
switching the list to another fuel silently re-targeted geofences. Now the
Alerts tab has its own 'Fuel to monitor' picker (persisted as
fuelboard.alertsFuel):
- Monitor reads loadAlertsFuel() on init/background relaunch
- Changing the alerts fuel re-registers geofences around stations selling
that fuel only
- Favourites priority slots stay scoped to the monitored fuel
- Tests: alerts fuel round-trip + default (33/33 passing)
